A Solo Founder Built an AI Scanner to Find SaaS Worth Cloning. The Best Answer Was 'Don't.'

The premise everyone already believes

Building SaaS used to require a team. A backend dev, a frontend dev, maybe a designer, someone to write the docs. Now a solo founder with the right agent stack can ship a functional product in a weekend.

That's not a hot take anymore — it's table stakes. The real question is: which SaaS is actually worth cloning?

Peter Vijeh asked that question and built an AI-powered scanner to answer it systematically. He fed it legacy software categories, had it score each one across multiple dimensions, and waited for the list of obvious targets. Instead, the most honest output the system produced was a warning: *don't*.

What the scanner actually measured

Vijeh's scanner wasn't vibes-driven. It scored SaaS categories across dimensions that matter specifically to a solo founder — not a funded team with a sales floor.

Two of the most important signals:

• `gtmFeasibility` (1–5): Can a solo founder realistically reach these customers without a sales team?
• `marketWorth` (1–5): Is the niche generating enough revenue to justify the attack?

This pairing is the whole insight. Plenty of legacy SaaS sits in markets with real revenue — but getting to those customers requires enterprise sales cycles, procurement committees, and relationships built over years. A solo founder with no sales team can build the product; they can't always build the distribution.

The scanner surfaced this gap category after category. High market worth, low GTM feasibility. The inverse showed up too: niches easy to reach, not worth the revenue.

Why the acquirer's playbook is the wrong map

Vijeh makes an observation worth sitting with: the criteria a PE acquirer uses to value legacy SaaS is almost exactly the inverse of what a solo founder should use.

Acquirers love sticky, high-switching-cost enterprise tools with long contracts and embedded workflows. Those are precisely the categories a solo founder should avoid — not because the product is hard to build (AI has largely solved that), but because the customer is nearly impossible to reach and convert without institutional sales infrastructure.

The solo founder's edge is speed, low overhead, and the ability to serve markets too small for a funded competitor to bother with. That edge disappears the moment you pick a category where sales cycles run six months and procurement approval requires a vendor security review.

This is the core reframe: build for the customer you can reach, not the market that looks biggest on a spreadsheet.

The categories that actually scored well

The scanner didn't come back empty. Certain categories threaded the needle — real revenue, reachable customers, legacy tools with bloated interfaces and pricing that hasn't moved in a decade.

The patterns in the high-scoring categories tend to share a few traits:

• Self-serve onboarding is the default. Customers in these niches expect to try before they buy, without talking to anyone.
• The incumbent is old. We're talking 2008-era UX, per-seat pricing that punishes growth, and a feature list built by committee.
• The buyer is a practitioner, not a procurement team. A single person with a credit card makes the call.
• Switching cost is real but not paralyzing. There's friction in moving, but a meaningfully better product at 40% of the price clears it.

Think niche project management tools for specific verticals, reporting layers on top of underserved data sources, workflow automation for professional services categories that Zapier never properly covered.

What AI actually changed (and what it didn't)

The received wisdom is that AI collapsed the cost of building. That's true. A solo founder can now ship what used to require a five-person engineering team.

What AI hasn't changed: distribution still requires finding people who have the problem, and convincing them your solution is worth switching for. The scanner's `gtmFeasibility` score exists precisely because building is no longer the bottleneck — getting to customers is.

This is why the one-person unicorn model works best when the founder picks markets where they can run distribution as a system, not a team. SEO, product-led growth, community, niche content — these are channels one person can operate. Cold outbound at scale, enterprise sales, channel partnerships — those need bodies.

The scanner's "don't" verdict wasn't about difficulty of building. It was about structural mismatches between the market and the solo founder's actual distribution capacity.

The metrics that make a solo SaaS worth cloning

If you're running your own version of this analysis, the numbers to anchor on:

• Revenue per customer: Niches where ACV is $500–$5,000/year are often the sweet spot for solo founders. High enough to matter, low enough that procurement isn't involved.
• Market size: You don't need a billion-dollar TAM. A $20M niche where you can capture 5–10% is a strong revenue-per-employee outcome when you're the only employee.
• Incumbent NPS signal: Check reviews on G2 and Capterra. If the top complaints are UI, pricing, and slow support — not missing features — that's a clonable product.
• Self-serve conversion benchmarks: Comparable PLG tools in the niche converting free-to-paid at 3–8% signal a market that will buy without hand-holding.

Build the scanner, read the output honestly

The most useful thing about Vijeh's project isn't the specific category scores — it's the methodology. Systematizing the decision of what to build is itself an AI-native behavior.

Most founders pick markets based on frustration (something that annoyed them), adjacency (something close to their last job), or trend-chasing (something that showed up in three newsletters). Running a structured scoring model against real market data before writing a line of code is the kind of leverage that separates intentional builders from lucky ones.

For founders working through how to build a one-person startup with AI, the scanner framework translates directly: score your candidates on GTM feasibility first, market worth second. If the first number is low, the second number doesn't save you.

The "don't" output is the most valuable thing the scanner produced. Most market analysis tools are optimized to tell you yes. A tool calibrated to tell you which opportunities are structurally wrong for your specific constraints is rare — and honest.

The AI made building cheaper. It didn't make bad market selection cheaper to recover from.

---

Is your company eligible? Submit to the leaderboard → onepersonunicorn.co/submit